The Three Critiques and

the Opus postumum

E C K A R T F Ö R S T E R
Editor

STANFORD UNIVERSITY PRESS

StanfordCalifornia
 Kanfs Notion of a Deduction and
the Methodological Background
of the First Critique

D I E T E R H E N R I C H

How did Kant conceive the program and die method of the tran-
scendental deduction in the first Critique? In trying to provide an an-
swer to this question, I shall draw on sources that have been hitherto
unused or unknown. At the same time, I shall point toward a frame-
work that could accommodate and account for all deductions in
Kant's work.
The range of problems and sources implicated is too extensive,
however, to be covered adequately within a single paper. Thus my
remarks will take the modest form of a research report. But as I pro-
gress I hope it will become apparent that a survey of the assumptions
and the context within which Kant's program of philosophical de-
ductions unfolds has a greater interest than the careful elaboration of
one of its facets. I shall, however, clarify Kant's notion of a philo-
sophical deduction in a more detailed fashion, and then turn to a
more summary treatment of further aspects of, and perspectives on,
Kant's philosophical methodology.
In spite of long-lasting efforts, the key chapter of the first Critique
resists penetration. We have no commentary that explains, in terms
of shared principles and intentions, the argumentative strategy Kant
employs—both in detail and as a whole, and for both editions jointly,
furthermore, attempts to reconstruct or to extend Kant's way of rea-
soning by means of similar but independent philosophical theories
 or forms of analysis cannot yet account convincingly for the nature
and the source of the similarities and differences between their own
projects and Kanfs. This situation appears not to be accidental. To
understand it, we must clarify the background against which Kant's
reasoning in the deductions unfolds. We have reason to believe fea-
tures of this background that Kant could take for granted have es-
caped us thus far.
Any successful interpretation of Kant's deductions must meet a
number of criteria, of which 1 shall mention only three.
First, an interpretation should be capable of comprehensively ex-
plaining the vocabulary that Kant employs when he comments on
the program of his deductions, in a way that exhibits the unity and
the internal connectedness of its various terms.
Sccond, an interpretation should provide means for understand-
ing the way in which Kant composed the texts of his deductions—
not only the two deductions of the first Critique, but also the deduc-
tions in Kant's work in general. The deduction of the first Critique,
to be sure, deserves special attention, for it was within the context of
this deduction that Kant became convinced that the program he car-
ried out under the name "deduction" is indispensable and is best
characterized by the term "deduction." Furthermore, the deductions
of the first Critique are by far the most extensive ones, and they stand
out clearly against the rest of the corpus of the first Critique because
of their style and the condensed and intense reasoning in which they
engage.
Third, an interpretation of Kant's deductive strategy must be able
to apply to the deduction of the second Critique. There the deduc-
tion depends upon a so-called fact of reason (which Hegel derided as
"the undigested log in the stomach, a revelation given to reason").
We are inclined to adopt, almost automatically, an understanding of
the term "deduction" that immediately creates a tension between a
deduction and reference to a fact. This results in misunderstanding
of Kant's argument in the second Critique—misunderstanding also
of the way in which it is sytematically related to the deduction of the
first Critique.

I
In Kanfs philosophical language, the meaning of the term "de-
duction" is different from what we almost irresistibly expect, and this
largely accounts for a persistent failure to understand Kant's deduc-
tions as a unitary and (within its unity) well-structured program. For
"deduction" is a term that is quite familiar to us. It refers to the logi-
cal procedure by means of which a proposition—namely, the conclu-
sion—Is established through the formal relationship of other propo-
sitions, Its premises. Thus we take a deduction to be a syllogistic
proof. Kant was familiar with this usage of the term "deduction."
Yet, unlike now, this was not the only, and not the most common,
usage in eighteenth-century academic language. Indeed, if we as-
sume that Kant announces under the heading of "deduction" a well-
formed chain of syllogisms, we must arrive at a very unfavorable con-
clusion about his capability of carrying out such a program. The
deduction of the first Critique, to be sure, does claim to be a proof.
But if it Is to be defined as a deduction on the basis of its correctness
and, above all, its perspicuity as a chain of syllogisms, its failure to
meet its own standards would be completely obvious. We know,
however, that within the first Critique Kant repeatedly showed his
ability to design accurate syllogistic proofs—for Instance, in the Re-
futation of Idealism and in the Antinomies. We have, consequently,
good reason to look for a reading of the term "deduction" in Kant's
sense, one that does not make the meaning of his very program en-
tirely dependent upon the design of a chain of syllogisms.
The literal meaning of "to deduce" (in Latin) Is: "to carry some-
thing forth to something else." In this very general sense it is not
restricted to derivations within a discourse, for one "deduces" a river
by digging a new riverbed. Within die range of the methods of dis-
course, "deduction" can have a variety of applications. A "deduc-
tion" in the original (Latin) sense can take place wherever something
results from a methodological derivation from something else.1 Rel-
ics of this very widespread usage are preserved in the European lan-
guages in various ways: in English, for instance, in "tax deduction."
Everybody who is familiar with the first Critique will remember
the first sentence under the heading "The Principles of Any Tran-
scendental Deduction": "Jurists, when speaking of rights and claims,
distinguish in a legal action the question of right {quid juris) from
the question of fact {quid,facti); and they demand that both be proved.
Proof of the former, which has to state the right or the legal claim,
they entide the deduction" (A84/B116). Since here Kant apparently
wants primarily to distinguish the two questions, one can easily be
led to the opinion that he employs "deduction" in the ordinary logi-
cal sense, with the further qualification that the premises of the syl-
logism should be capable of justifying legal claims—thus, presum-
ably, normative propositions.
But by adopting this seemingly natural and almost irresistible
reading one has already missed what is distinctive to the methodo-
logical idea that gives to Kant's deductions a unitary structure. One
also misses the reasons for which Kant refers to the juridical para-
digm, and why he could and did structure the first Critique in its en-
tirety around constant reference to juridical procedures. In four steps
I shall try to elucidate the background of Kant's adoption of the term
"deduction" from its juridical context and set out the reasons why he
transferred it to his philosophical program.
i. By the end of die fourteenth century, there had come into
being a type of publication that by the beginning of the eighteenth
century (when it had come into widespread use) was known as De-
duktionsschriften ("deduction writings5'). Their aim was to justify
controversial legal claims between the numerous rulers of the inde-
pendent territories, city republics, and other constituents of the
Holy Roman Empire. They presupposed both the invention of print-
ing and the establishment and universal recognition of the Imperial
Courts as an authority above the otherwise largely independent
members of the Empire. These deduction writings were not sold by
publishing houses but were distributed by governments, with the in-
tention of convincing other governments of the rightfulness of their
own positions in controversies that might eventually lead to military
force and thus to the need of finding support from other rulers. Be-
fore the final decisions of one of the Imperial Courts (which were by
no means always respected), legal proceedings also required that a
deduction had to be submitted by both parties. Most of the legal
controversies concerned inheritance of territories, the legal succes-
sion in reigns, and so forth. In all the cases extensive arguments
about the way in which a claim had originated and had been main-
tained over generations had to be given.
The size of the deduction writings varied, from brochures to folio
volumes of up to three thousand pages. Governments preserved de-
ductions in special libraries so that they could be drawn upon in
unforeseeable future conflicts. Since their printing was often very
fine and elaborate, deductions were also collected—for instance, by
former diplomats, who had access to them. Thus, after centuries,
auctions of special deduction collections could bring considerable
amounts of money. Approximately twelve thousand deductions were
published from the fifteenth century to the eighteenth century. At
the beginning of the eighteenth century, it proved profitable to pub-
lish bibliographies of the deductions. Writing deductions was a
juridical specialty, and a famous deduction writer could easily be-
come rich. The most admired deduction writer of Kant's time was
J. S. Putter, professor of law at Güttingen and coauthor of die text-
book that Kant used in his frequent lectures on natural law.2
It can be shown that Kant was familiar with the practice of deduc-
tion writing. He was for six years a librarian of the royal library in
Königsberg and had to check its stock when he assumed his office.
He uses the terminology of the deductions in his own little juridical
quarrels (C 12:380, 421). And he speaks occasionally about deduc-
tions in the archives that had been ignored by rulers who preferred
to use violence (MM §61, 6:350). Kant also had very good reasons to
assume—given die widespread practice of arguing through deduc-
tion writings—that his audience would understand him when he
transferred the term "deduction" from its juridical usage to a new,
philosophical one. What he could not foresee was that such a wide-
spread usage would very soon become obsolete, when the Holy Ro-
man Empire was abolished under pressure from Napoleon. With
this, the Imperial Courts and the practice of writing deductions dis-
appeared forever, and the term "deduction" became extinct and al-
most incomprehensible. With regard to the Critique and its deduc-
tions, we can thus understand in a new light die old saying that
books, too, have their destiny.
2. The practice of deductions reaches back to a time when the tra-
dition of Roman law was not yet revitalized and the modern theory
of law had not yet been founded. These two processes resulted in a
need to refine and regulate the practice of the writing of deductions,
and this affected Kant's way of conceiving their philosophical corre-
late in the first Critique. The old deductions were perceived by die
new generations of jurists as clumsy and largely unfit for the pur-
poses for which they were written. Thus, within the fast-growing
methodological literature on law, academic jurists provided analyses
of what a deduction was and guidelines for a deduction's author.
This literature provides clues for reading the transcendental deduc-
tion as well. In 1752, for instance, one of the methodologists pro-
duced the following criteria for a good deduction: Since a deduction
is not a theory for its own sake, but rather an argumentation in-
tended to justify convincingly a claim about the legitimacy of a pos-
session or a usage, it should refrain from unnecessary digression,
generalizations, debates about principles, and so forth, which are of
interest only to the theoretician. A deduction should be brief, solid
but not subde, and perspicuous.
The author of this treatise praises highly a practice of Pütter, the
famous deduction writer. If Pütter did not succeed in producing a
deduction that satisfied these criteria, he would write-a second de-
duction, a brief and elegant text that summarized the main points of
his argument. I have checked some of Pütter's deductions, and there
is indeed one to which such an additional text is attached, printed in
different format and on less expensive paper; it carries the title "Brief
Outline {Kurzer Begriff ] of the Zedwitz Case." Now, if one looks at
the text of Kant's transcendental deduction, one finds at the end of
it a similar summary, which Kant gives nowhere else. More impor-
tantly, the summary carries in the second edition the very same tide:
"Brief Outline [Kurzer Begriff ] of this Deduction" (B168).
It does not seem very probable that this is an accident (although
the phrase Kurzer Begriff as tide for an outline was indeed common
in eighteenth century scholarly language). It appears that Kant
thought his deduction as a text should be modeled on the juridical
paradigm and meet its criteria of excellence. Thus we conclude—be-
fore analyzing the argumentative form of the deduction—that Kant
wrote the text of the deduction of the first Critique following the
standards of a good juridical deduction, which focuses exclusively
upon justifying a claim. Not only is the argumentation of the deduc-
tion correlative to the juridical argumentative form (as we shall see
shortly), the text of the transcendental deduction is a deduction writ-
ing in the technical sense. That explains why Kant could not agree
with those who complained that he refrains from extensive theoriz-
ing. He deliberately intended to be brief and to focus exclusively
upon his crucial points. In the first edition of the deduction he says
explicitly that "he intends to avoid elaborate theory. Later he recom-
mended the second-edition deduction because it arrived at the in-
tended result by the easiest way.
We must now turn to questions about the argumentative form
of a juridical deduction. They have been discussed by the theoreti-
cians of natural law, and the first to come up with a definition of
what a deduction consists in was Christian Wolff. The basic distinc-
tion between types of rights is between innate and acquired rights.
In J. S. Pütter and G. Achenwall (the authors of Kant's textbook),
these are called absolute and hypothetical rights, respectively. Hypo-
thetical rights originate in a "fact" (factum, meaning both "fact" and
"action"), which must exist before the right In question can come
into being—mostly from an action by virtue of which die right is
"acquired." Innate or absolute rights, conversely, are Inseparable
from a human being as such. Humans by their very nature possess
such rights.
But acquired rights have a particular origin. I have the right, for
example, to carry a title of nobility If I am the legitimate child of a
particular couple. I have the right to carry an academic title if I have
successfully passed the examinations without fraud. I can use a par-
ticular good—a house, for instance—If I have purchased it by a valid
contract or if I have Inherited it by a valid last will.
In order to decide whether an acquired right is real or only pre-
sumption, one must legally trace the possession somebody claims
back to its origin. The process through which a possession or a usage
is accounted for by explaining its origin, such diat the rightfulness of
the possession or the usage becomes apparent, defines the deduction.
Only with regard to acquired rights can a deduction be given. This
implies that by definition a deduction must refer to an origin.
We now understand why these two notions, the methodologi-
cal notion of a deduction and the epistemological notion of an origin
of our knowledge, are inseparably linked in the terminology of the
first Critique. In this context the question that Kant raises constantly
in the Critique also exhibits its distinctive meaning: "How is it pos-
sible . . The question does not ask for one or anodier sufficient
condition for our possession of knowledge. In a state of doubt about
the rightfulness of our claim to be in die possession of genuine
knowledge, it seeks to discover and to examine the real origin of our
claim and with that the source of its legitimacy.
But doesn't this confuse and blur the distinction between explana-
tion and validation, between the question of right and the question
of fact, to which Kant assigns such importance throughout the open-
ing paragraphs of his first deduction? In response to this question we
can observe that the distinction between the two questions (of right
and of fact) cannot be drawn in such a way that only the question of
fact remains concerned with the origins of our knowledge. Both
questions require an understanding of the origin, but each in its par-
ticular way.
 Consider the example of a last will: It is possible to tell a story
about the way in which the will has been conceived of and arrived at,
when it was written and how it has been preserved. This is what the
deduction writings call Geschichtserzählimg ("report of the story35),
or speciesfacti. Such a speciesfacti can be produced in court and can be
disputed—for instance, if it becomes doubtful whether a possession
or a usage exists at all. But it cannot setde by itself the quaestio juris.
To answer this question, one has to focus exclusively upon those as-
pects of the acquisition of an allegedly rightful possession by virtue
of which a right has been bestowed, such that the possession has be-
come a property.
It should be mentioned in passing that the idea of an acquisition
of legal tides does not necessarily presuppose a particular legal sys-
tem with reference to which the entitlement becomes decidable. The
Natural Right Kant uses as his paradigm recognizes an original ac-
quisition. The conditions of its rightfulness can be determined prior
to any particular legal system. The categories of the pure under-
standing are justified with regard to their "original acquisition55
through an operation of the mind by means of Kant's deduction of
the categories within the first Critique. It is important to realize, fur-
thermore, that the quaestio juris can be answered in a satisfactory way
even if the quaestio facti meets with insurmountable difficulties. Con-
sider again the example of the last will: in many cases we are unable
to produce a complete story of the way in which the will has been
made. But if it can be determined in court that the will is authentic
and valid, by means of only a few but crucial aspects, the question of
right can still be answered decisively.
This consideration can be applied to the transcendental deduction
of the categories. Kant is of the opinion that it is impossible to
produce an adequate species facti about the acquisition of our knowl-
edge. The text appears to suggest that the story about acquisition
that Locke and others had to tell is possible, albeit irrelevant. How-
ever, from other sources we can clearly show that Kant's position was
a different one. Within philosophy, to the species facti of the jurists—
the report of the story—corresponds what Kant calls the "physiol-
ogy of reason." For a number of reasons he became convinced that
such a physiological account is impossible. For Kant, Leibniz as well
as Locke is a physiologist of reason. This description implies a two-
fold criticism on Kant5s part, (i) The attempt of these philosophers
to give a complete account of the roots and the genesis of our ration-
ality is not a promising undertaking. (2) They also refrain from
doing what ultimately matters in philosophy: justifying the claims of
reason against skepticism. Critical philosophy thus opens an entirely
new pathway, which can be defined in terms of what the notion of a
deduction implies, once it is understood in its distinctive sense.
But deductions can never be given without reference to the facts
from which our knowledge originates. We cannot arrive at, and
don't need a comprehensive understanding of, the genesis and con-
stitution of these facts in themselves. Yet we must arrive at an under-
standing of the aspects of them that suffice to justify the claims at-
tached to our knowledge. Most of the facts the deductions rely upon
are basic operations of our reason. The deductions refer to the intrin-
sic, quasi-Cartesian forms of these operations: their independence
from particular experiences. Yet their status as operations and forms
of operations does not exhaustively define their roles as principles
upon which a deduction can be built. Operations that are facta
(therefore actions in the juridical sense) imply factual elements that
cannot be explained by virtue of actions we can always perform.
Most of the origins from which Kant's deductions are derived ex-
hibit clearly this additional factual element. The features shared by
the unity of apperception, the consciousness of space and time as
such, and the moral law as a fact of reason illustrate this common
feature of the principles according to which Kant's deductions have
been designed. Viewed in this light, the deduction of the second Cri-
tique does not deviate from the general pattern Kantian deductions
exhibit, whatever their particular course.
The differences between Kant's deductions can be explained by
means of the different ways in which we have access to the origins
and the principles of our discourses, and by variants in the notion of
an origin itself. These differences result in the distinction between
strong and weak versions of philosophical deductions, which 1 have
discussed elsewhere.3
4. We don't know just when Kant decided to present his new
method of philosophical justification in the form and. the termi-
nology of juridical deductions. This probably happened rather late in
his long preparations for the publication of the first Critique. We do
know, however, that the decision concerned more than the chapter
that became entitled "Transcendental Deduction of the Pure Con-
cepts of the Understanding." The fact that Kant composed this chap-
ter in the style of a deduction writing largely explains its unique
character, which contrasts markedly with the rest of the first Critique.
But the entire first Critique,j and the way In which Kant presents its
theory as a whole, was thoroughly affected by the decision to adopt
juridical procedures as a methodological paradigm.
The Critique Is not just permeated by juridical metaphors and ter-
minology. Its major doctrines are related to one another by means of
the theory of legal disputes presented by Pütter and AchenwalL A
legal dispute originates when a party's claim has been challenged by
an opponent, so that a court case must be opened. This happened In
philosophy when the skeptic challenged the claim of reason to be
in possession of a priori knowledge of objects. The dispute makes
indispensable an Investigation into the origins of such knowledge.
To the extent to which a deduction can be produced, the claim of
reason becomes definitely justified and the challenge of the skeptic is
rejected. This is the aim of the Transcendental Analytic.
But there is the other possibility, to which the Transcendental
Dialectic corresponds: a deduction might turn out to be Impossible.
If the claim to knowledge that transcends the limits of experience
cannot be justified, the challenged party must retract its claim. But
that does not necessarily mean that the opposing party wins the
court case. For the skeptic—here, in the guise of the empiricist—
may in turn be unable to validate his claim that the usage of Ideas
transcending experience is Illegitimate and an empty presumption.
In such a situation, where conflicting claims cannot be settled in
court in favor of one of the parties, the parties threaten to become
engaged in an endless struggle, which would destroy the peace and
result in a war within reason itself. The ruling of the court of reason
in such a situation consists In an order to keep the peace. The dog-
matic philosopher must refrain from claiming that he is in possession
of valid knowledge beyond narrow limits. "A complete review of all
the powers of reason—and the conviction thereby obtained of the
certainty of its claims to a modest territory, as also of the vanity of
higher pretensions—puts an end to the conflict, and Induces It to
rest satisfied with a limited but undisputed territory55 (A768/B796).
This Is, of course, the territory whose borders are drawn by the con-
ditions of possible experience. But the rule of the court to keep the
peace does not permit the challenger to deny our right to use Ideas of
reason that transcend experience. Since neither a deduction nor a
proof of the emptiness of these Ideas could be produced, and since it
has been shown that both proofs are impossible, reason, which Is in
possession of these ideas, becomes entitled to use them as long as it
refrains from claiming that they are used as justifiable knowledge.
The court of reason rules on the basis of the juridical principle that
applies in such cases: if a dispute about the rightness of a usage can-
not be settled, the usage remains with the possessor: "melior est con-
ditio possedendi" (A777/B805). Because reason remains in possession
of its ideas, though in using them it cannot lay claim to any knowl-
edge, the way to a purely practical philosophy remains open.

This clarification of the precise meaning of the term "deduction"

in Kanfs work makes possible two conclusions about the argumen-
tative structure of the transcendental deduction of the categories.
First, in its fundamental structure the transcendental deduction is
patterned on a deduction that aims to justify an acquired right by
appealing to particular features of the origin of the categories and
their usage—features that had been challenged. Different parts or
steps discernible within the text of the deduction can be explained
primarily as partial moves aimed to elucidate the origin of the usage
of the categories—and thus as a partial answer to the question about
the conditions that would make possible a legitimate usage of the
categories. These steps can also function as links within a chain of
syllogisms, but such functioning by itself does not make them partial
moves within a juridical deduction. This is particularly important to
our understanding of the structure of the deduction in the second
edition of the Critique. The deduction is indeed a proof, and it
brings its various partial results together by means of a syllogistic
chain, but its being a "deduction" is not defined in terms of a chain of
syllogisms. Any relatively independent part of it must also be a rela-
tively independent move in the uncovering of the origins of the
usage of the categories. For the purpose of the deduction is to deter-
mine, with regard to origin, the domain and the limits of the catego-
ries5 legitimate usage.4
Second, once we have realized that the deduction as such cannot
be founded upon a syllogistic structure, we have gained a new flexi-
bility in our understanding of the various kinds of argumentation
Kant might employ in the course of the deduction. The very notion
of a deduction is compatible with any kind of argumentation suitable
for reaching the goal—namely, the justification of our claims to a
priori knowledge. As a matter of fact, several types of argument op-
erate within the text of the deduction before it begins to establish its
results by means of a syllogistic proof The task of a philosophical
commentary—as vet unwritten—would be to discriminate these
j J

types and to assess their respective function and philosophical value.

Within such a context, we could reopen the question about whether
Kant employs a particular kind of argument (that is, in some sense,
distinctly "transcendental55).

II
But further problems must first be taken up. Although we now
understand the program implied bv the notion of a deduction, we
don't yet know anything about Kant's ideas concerning how to ar-
rive at a philosophical deduction. We must still explore his views
about the methodological foundations on which one might justify
acquired rights in philosophy.
In this regard, the first Critique remains completely silent. It em-
ploys a number of methodological terms: it examines origins; it in-
spects reason as such; it searches for sources and explores how we can
proceed from them; it explains possibilities; it investigates content,
usage, and right; and it designs proofs that indicate conditions of
possibility. But it does not analyze or account for any of these terms.
It is not unusual in philosophy for completely new theories to be
unable to explain their own procedures. The terms Kant employs in-
dicate a complex and evasive field of problems, and so he probably
had good reason to concentrate on the content rather than on the
methodology of his project. It would, however, be disappointing if
nothing could be said about Kant's methodological assumptions and
about his theory of the foundations underlying the practice of de-
ducing knowledge claims.
To pursue this topic, one must turn to yet another literature. This
body of texts—the literature of applied logic—has been forgotten
almost as completely as the literature of juridical deductions. Kant
covered its problems extensively in his lectures on logic (although he
excludes these problems from what belongs to logic proper in a nar-
row sense). There he sketched his views about philosophical meth-
odology, including the methodology of the Critique. I shall present
Kant's views in four more steps.
i. The first step in examining Kant's views on methodology is to
note a few aspects of his assessment of proofs within philosophy. In
assessing the role of syllogisms in philosophy, Kant follows Descartes
and the school of Rüdiger: syllogisms are secondary, merely sub-
sequent organizations of knowledge already acquired. What matters
most in philosophy is to ascertain the reliability of premises (of
Beweisgründe, that is, of the notions and reasons that proofs can
rely upon).
He also believed that philosophical knowledge cannot be based
upon logical possibility. It must find what Kant calls "real reasons'5
(rationes verae). Using them, it must show in what way knowledge
issues from its real source. In this sense, philosophy has to provide
"genetic55 explanations. (This corresponds to what die deduction
provides—sources or origins.)
Such explanations, for Kant, could never become "demonstra-
tions.55 They are "probations55 (probationer). Demonstrations are pos-
sible only in mathematics. Mathematical demonstrations provide os-
tensive knowledge, but philosophical knowledge cannot become
equally secure. It is always possible that in the course of philosophi-
cal reasoning one might overlook an important aspect of the prob-
lem. For that reason, philosophical argumentation must be holistic
in the following sense: any result we arrive at must be checked against
results that have been gained in other fields of philosophy. One can-
not present philosophical insights without hesitation, and separately
from die other proofs we are inclined to accept (see, for example,
R2513, 16:400). This stance explains Kant's claim that the Critique
becomes convincing only by virtue of the totality of its theorems and
proofs. It explains, furthermore, why the deduction is not situated at
the beginning of the work (where one might expect it). It also ex-
plains why Kant does not try to achieve in the deduction the per-
spicuity and the rigor distinctive to the kinds of knowledge it is
about to justify. Justification as a method cannot excel in the forms
of discourse for whose sake the justification is undertaken, and it
cannot compete with the clarity and the perspicuity of foundational
disciplines as they had frequently been conceived within the tradition
founded by Leibniz. But these disciplines cannot really solve—can-
not even address—basic philosophical problems, and for that reason
they are in a sense unphilosophical."
2. But these three theorems taken together cannot by themselves
clarify the epistemic basis upon which a deduction has to rely. For
this, we must take a second step and consider a central and basic dis-
tinction Kant frequendy makes in his lectures beginning in the early
1780's (see 24:161 ("Logik Blomberg"), 424,547, 641), which is pres-
 ent also in the first Critique (although there one can hardly grasp its
methodological importance). The distinction Is between "reflection"
(Überlegen, reflexio) and "Investigation" (Untersuchen, cxaminatio).
The Critique Is an examination, or Investigation. Since Kant claims
that reflection precedes Investigation, It Is plausible to suppose that
reflection Is the source by means of which an investigation can be
undertaken. Kantfs theory of reflection (which Is utterly differ-
ent from the meaning of "reflection" that became current in post-
Kantian philosophy) Is as follows, (a) Our cognitive- capacities are a
"mingled web." They cannot be reduced to one single form of funda-
mental Intelligent operation, (b) Each of these capacities becomes
operative spontaneously and with regard to its appropriate domain,
(c) To arrive at genuine knowledge, It is necessary to control and to
stabilize these operations and to keep them within the limits of their
proper domains. Our mind must regulate when a particular activity
comes into play and be sure that it alone remains operative. For that
purpose, the mind must Implicitly know what is specific to each of Its
particular activities. This implies, furthermore, that the principles
upon which an activity is founded must be known by contrast with
the other activities. Reflection consists In precisely this knowledge.
Without it we would, for example, confuse counting with calculat-
ing, analysis with composition, and so forth. Kant says explicitly that
without reflection we could only utter meaningless sequences of
words, (d) Therefore, reflection always takes place. Without any
effort on our part, we always spontaneously know (albeit, informally
and without explicit articulation) about our cognitive activities and
about the principles and rules they depend upon. Reflection in this
sense is a precondition of rationality.
Reflection is not introspection. It accompanies operations inter-
nally. It is not the achievement of a philosopher who, by means of a
deliberate effort and within an intentio obliqua., turns inward to ex-
amine the operations of reason. Thus it is a source, not an achieve-
ment, of philosophical Insight. Now notice the similarity of and con-
nection between "reflection" on the one hand and the program of a
"deduction" on the other:

Deductions are founded upon a partial knowledge of significant fea-

tures of the origin from which our knowledge arises.
Reflection is not a descriptive, let alone an exhaustive knowledge of
the processes and operations of cognition. It is only an awareness
 of what is specific to them, presumably the general principles and
rules upon which they rely.

Thus it appears that deductions can be built upon a reflective knowl-

edge of precisely this kind.
3. We take a third step when we note that the Critique—and. with
it, its deductions—is an investigation by means of which claims
about knowledge are examined. Kant defines investigation and ex-
amination as correlates of reflection. We reflect always, but investiga-
tion is a deliberate activity. It is only undertaken when doubts about
and challenges to knowledge claims have arisen. Then we must
search for the ground upon which our (real or only presumptive)
knowledge is founded—eventually we must try to produce a "deduc-
tion." But the investigation cannot depart from the domain widiin
which reflection is operative: it detects connections of which reflec-
tion itself is not explicitly aware. And it relates the principles that
orient a discourse to fundamental facts and operations that consti-
tute it yet which can also interpret and validate it.
These are the very facts we referred to when we clarified the philo-
sophical correlate of the juridical deduction—notably, the unity of
apperception, space and time, and the fact of reason. In exceptional
cases deductions might have to transcend the limits of the domain
that is disclosed in reflection: for example, in the case of the deduc-
tion of the reality of freedom. But even then they rely upon prin-
ciples and fundamental facts we already know about by reflection,
although we understand them and their central position within the
discourse in question only by means of an investigation. The system-
atic interconnectedness of the various forms of discourse can also be
understood by means of investigation. But investigation is preceded
by, and made possible through, reflection, by which the multidimen-
sional system of our cognitive capacities is accessible to us, persis-
tently and prephiiosophically.5
Two corollaria can be added to this third step:
(a) Since the deduction, as an investigation, always depends in-
ternally on what reflection provides, we can reasonably expect that
no deduction can get under way unless it relies primarily on argu-
ments that refer direcdy to what is revealed by reflection. These argu-
ments constitute the core of every transcendental deduction. And
their formal feature is a clarification of the awareness that a particular
operation cannot be carried out unless another and more fiindamen-
tal operation comes into play. This is the distinctive feature of the
arguments that appear within the Critique in the grammatical form
"not without." To this type belongs the argument in the second-edi-
tion deduction that analysis is not fundamental but is always intrin-
sically accompanied by synthesis on a deeper level, and the argument
of the first-edition deduction to the effect that synthesis in turn re-
quires principles of unity that cannot originate within experience.
Kant appears to believe that the key argument of the deduction con-
necting the unity of apperception with the principles of unity that
guide all synthesis belongs also entirely to the same type. But in my
opinion this is true only with qualifications and must be further ex-
plored. But such an exploration presupposes that the methodology
of a deduction based upon reflection has been understood.
(b) We can indicate a reason for Kant's reluctance to present ex-
plicidy his philosophical methodology. (Almost all of the above has
been taken from transcripts of Kanfs lectures; very little of it is in his
printed texts. He must have had reason for being reticent.) Since re-
flection is a permanent, albeit implicit, knowledge, and investigation
is a deliberate undertaking on the part of the philosopher, there re-
mains a gap between these two cognitive activities, regardless of the
essential correlation between them. Thus the question arises as to
how an implicit knowledge can be transformed into an explicit one.
There must be a device by which the transition can be made in a
methodologically safe and respectable way. Kant realized this and
was inclined to put the theory of "preliminary judgments" ( judicia
praevia) to work at precisely this place: within reflection a tendency
to conceptualize our cognitive faculties in a particular way somehow
arises. These "preliminary judgments" are the point of departure for
philosophical investigation. The investigation does not have to ac-
cept them, but it starts off from them. In his lectures on logic, above
all in the Vienna transcript, Kant acknowledges that we have only a
rudimentary understanding of this mechanism and that for this rea-
son a satisfactory methodology of philosophical reasoning meets
with hitherto insurmountable difficulties.
4. In a fourth and last step we turn to an important and surpris-
ing application of K a n f s doctrine of the role of reflection within phi-
losophy. The key notion of the deduction in the first Critique is,
without doubt, the unity of apperception. Much can be said in favor
of the view that this principle must be relied on, at least indirectly, in
the other transcendental deductions as well. Now, when Kant turns
to discuss this principle, he refers to it constantly as "die I think."
Presumably a number of reasons are responsible for the persistent
usage of this peculiar phrase. But among them is one that can be di-
rectly derived from Kant's theory of reflection: the awareness " I
think" is precisely the self-consciousness that can be attached to
natural and spontaneous reflection. And it is, in addition, the self-
consciousness that can accompany every kind of reflection, regardless
of the field of its employment. We can see this if we consider that: (a)
It is neither concept nor intuition and does not belong to any of the
various cognitive activities, (b) It is established prior to all kinds of
theorizing, (c) It emerges from an operation. But this operation is
not itself an act of reflection nor does it define reflection as such, (d)
Yet it potentially accompanies every case of reflection and is not re-
stricted to a specific area of reflective awareness or a particular dis-
course whose principles are disclosed by virtue of reflection. It has
the same generality and scope as reflection, and can thus be thought
together with any act of reflection. It is, as Kant says, "The I as sub-
ject of thinking, that is to say the pure apperception, the reflected I"
(Anthr §4, 7:i34n).6
It is impossible to understand diese passages without the help of
Kant's notion of reflection in its distinctive sense. But once the no-
tion of reflection has been clarified, we can also understand the fun-
damental role the unity of apperception plays within Kant's system
as a whole. The ultimate justification of the principles of our knowl-
edge should depend upon an origin that has a central position within
our cognitive system as it is accessible to us by virtue of the implicit
knowledge of reflection. This suggests that the principle by which
the most fundamental deduction can be carried out will have the
generality and unrestricted applicability that is the distinctive feature
of the process of reflection upon which the method of philosophical
justification or investigation continuously relies.7 In this manner the
key notion of the deduction in the first Critique and the methodo-
logical principle of all philosophical deductions, the correlation of
reflection and investigation, appear to be related to each other.

This leads us to a final point. The unity of apperception has ap-

peared in two very different roles: On the one hand, it is an aware-
ness that can accompany any reflectively accessible knowledge. But
this knowledge is manifold and appears to lack systematic unity. On
the other hand, the unity of apperception is the origin of the system
of the categories and the point of departure for the deduction of the
legitimacy of their usage. It is easy to sense the tension between the
loose generality of reflection, on the one hand, and the rigidity of
the claim that issues from the program of the deduction, on the other.
The Critique claims that reason as such is a system and, furthermore,
that philosophy must provide an exhaustive account of its principles
and of the various modes of its usage.
The tension disappears once we consider reflection in a slightly
different perspective: Kant's theory of reflection is founded upon the
observation that our various discourses would become confused and
inconsistent if they were not accompanied and guarded by a constant
process of reflected control But we might wonder whether such a
confusion would be likely to occur if the discourses were not system-
atically related to one another, so that erroneous transitions can sug-
gest themselves. Within a reason that was nothing but a bundle of
independent activities, the constant danger of confusion could hardly
occur. Such a reason might function properly as a coordinated set of
cognitive machines.
But reflection is omnipresent because reason is one, in spite of its
comparatively independent operations. The unity of reason, as far as
the systematic structure of its principles is concerned, is represented
in the most fundamental way by the implications of the thought
"I think," the system of the categories. But the very same thought
is intimately and universally related to the process of reflection
upon which the methodology of philosophical justification must be
founded.
Deductions cannot assume the shape of a rigorous and exhaustive
reasoning. Yet the philosophical theory of reason whose argumen-
tative core they provide must be systematic and exhaustive. But the
discrepancy between loose argumentation on the one hand and a
lofty pretension about systematicity on the other turns out to be il-
lusory. The difference between the two programs is to be understood
as the difference between two tasks within one single program, which
Kant has conceived and designed in a perfectly consistent way.
 1. Lewis White Beck, Early German Philosophy (Cambridge, Mass., 1969),
pp. 463-67; idem, Studies in the Philosophy of Kant (Indianapolis, 1965),
pp. 56-60.
2. H. J. Vleeschauwer, La deductwn transcendentde dans Poeuwe de Kant,
I (1934; rpt. New York, 1976), p. 255.
3. I cannot accept Professor Carl's statement (this volume, page 6) that
"The 'intellectual representations' of the Dissertatio are explicitly excluded."
Where explicitly? Where, indeed, implicitly? They are explicitly mentioned
three times in the long first paragraph and seem to me to be the central topic
of the letter. When Kant denies the usus realis of the intellect with respect to
objects of the senses, he is saying that intellectual concepts have no simple
and direct relation to the objects of experience (see ID §5, 2:393).
4. This much is explicitly retained in the letter, where Kant says that con-
cepts of magnitude (but not of quality) of sensible objects can be determined
a priori (C 10:131, lines 25-26; in A. Zweig's translation, p. 72).
5. The Reflexionen cited in Carl's note 3 are of little help in deciding
whether Ding and Sache name noumenal or phenomenal objects. R4644
suggests that Realbegriffe (concepts in the usus realis of intellect) do not refer
to phenomenal objects; R4276, on the other hand, seems to require that
Sachen be objects of experience. In his Reflexionen Kant tried almost every-
thing, and therefore their evidential weight is slight except when they sus-
tain and clarify each other.
6. In Inaugural Dissertation §24 we were told not to ascribe any sensible
predicates to intellißihilia. Reflexion 4644 warns against the converse error,
which I understand to mean the use of the intellectual concepts of the Dis-
sertation as if they were categories in the sense of the Critique: "Wenn man
von Dingen, die man nur unter der Erscheinung des Raumes kennt, all-
gemeine realbegriffe des Verstandes braucht (von der Art, wodurch dem
Verstände die Sachen selbst gegeben werden und deren qualitaet, nicht
die Grösse oder Möglichkeit), so begeht man eine pet itkm ein noumenF
(17:623.6-10), which is the converse of the error warned against in the Dis-
sertation, here called petitw phaenomen ovum.
7. Norman Kemp Smith, A Commentary on Kanfs Critique of Pure Rea-
son, 2d ed. (London, 1923), p. 206.

h e n r i c h : Kant's Notion of a Deduction

1. An early example is the derivation of a notation for a scale in music
from certain natural qualities of the tones: "do-re-mi-fa. .
2. G. Achenwall and J. S. Piitter, lus naturae, appeared from 1750 011 in
numerous editions. The second volume is reprinted in the Academy edition
of Kant's works (vol. 19). From the third edition on, the work appeared only
under AchenwalTs name.
3. See Dieter Henrich, "Die Deduktion des Sittengesetzes, 5 5 In Denken im
Schatten des Nihilismus cd. A. Schwan (Darmstadt, 1975), pp. 55—56.
4. With this remark I modify part of my "The Proof Structure of the
Transcendental Deduction," Renew ofMetaphysics, 22 (1969): 640—59. When
I wrote the paper, I had no idea what a deduction consists in and took for
granted that it was exhaustively defined as a chain of syllogisms. But it Isn't,
and after finding out that this is so, I must relativize what I said in that
paper. The deduction of the second edition is indeed a proof within two
steps; but Kant's main reason for separating the two steps is their distinctive
contribution to an understanding of the origins of knowledge. This result is
compatible with the analysis of the logical relations between the conclusions
of the two steps that I gave i n 1969.
5. One could show that a clear .connection exists between Kant's claim
that philosophy is based on natural reflection and his affiliation to Rousseau,
In whose work the ordinär}7 man in a sense knows everything from the very
beginning.
6. A few pages later he refers to it as "the I of reflection.55
7. The deduction of the categories has still to be given for features of the
"I think55 that are not in focus when the general notion of reflection is dis-
cussed: Its quasi-Cartesian status and its relation to truth and the form of a
proposition as such.

G Ü T E R : Psychology and the Deduction

1. Translations from the Critique ofPure Reason are my own, based on the
text by Raymund Schmidt (Hamburg, 1926).
2. Although Locke also thinks little needs to be said about the last
of these.
3. Thomas Hobbes, De Corpore, eh. 1, art. 2, in Body, Man, and Citizen,
ed. R. S. Peters (New York, 1962), p. 24.
4. Thomas Hobbes, Human Nature, ch. 4, art. 10, in Peters, p. 194.
5. David Hume, A Treatise of Human Nature, ed. L. A. Selby-Bigge, 2d
ed. rev. P. H. Nidditch (Oxford, 1978), I, III, vi, p. 89.
6. See generally, Hume, Treatise, I, III, Hi, pp. 76-82, and I, III, vi,
pp. 86-89.
7. Ibid., I, III, Hi, p . 82. 8. Ibid., I, III, v i , p. 92.
9. Ibid., I, III, viii, p. 98. 10. Ibid., I, III, xiv, p. 156.
11. Ibid., p. 165. 12. Ibid., I, III, vi, p. 94.
13. Ibid., I, IV, I, p. 180. 14. Ibid., p. 181.
is. Ibid., p. 182. 16. Ibid.
17. Ibid., p. 183. 18. Ibid., p. 185.
19. Ibid., p. 183.